Water Research 177 (2020) 115767

Contents lists available at ScienceDirect

Water Research
journal homepage: www.elsevier.com/locate/watres

CN-China: Revised runoff curve number by using rainfall-runoff

events data in China
Huishu Lian a, b, Haw Yen c, Jr-Chuan Huang d, Qingyu Feng e, Lihuan Qin a,
Muhammad Amjad Bashir a, Shuxia Wu a, A-Xing Zhu f, g, Jiafa Luo h, Hongjie Di i,
Qiuliang Lei a, *, Hongbin Liu a, **
a
Key Laboratory of Nonpoint Source Pollution Control, Ministry of Agriculture, Institute of Agricultural Resources and Regional Planning, Chinese Academy
of Agricultural Sciences, Beijing, 10081, China
b
School of Environmental Science and Engineering, Southern University of Science and Technology, Shenzhen, Guangdong, 518055, China
c
Blackland Research and Extension Center, Texas A&M Agrilife Research, Texas A&M University, Temple, TX, 76502, USA
d
Department of Geography, National Taiwan University, Taipei, 10617, Taiwan
e
Research Center for Eco-Environmental Sciences, Chinese Academy of Sciences, Beijing, 100085, China
f
Key Laboratory of Virtual Geographic Environment, Nanjing Normal University, Ministry of Education, Nanjing, 210023, China
g
Department of Geography, University of Wisconsin-Madison, Madison, WI, 53706, USA
h
AgResearch Limited, Ruakura Research Centre, Hamilton, 3240, New Zealand
i
Centre for Soil and Environmental Research, Lincoln University, Lincoln, Christchurch, 7647, New Zealand

a r t i c l e i n f o a b s t r a c t

Article history: The curve number (CN) method developed by the United States Department of Agriculture (USDA) in
Received 22 September 2019 1954 is the most common adopted method to estimate surface runoff. For years, applicability of the CN
Received in revised form method is a conundrum when implementing to other countries. Specifically, countries with more
2 March 2020
complex natural environment may require more dedicated adjustments. Therefore, the current CN look-
Accepted 28 March 2020
Available online 2 April 2020
up table provided by USDA might not be appropriate and could be questionable to be applied directly to
regions elsewhere. Some studies have been conducted to modify CN values according to specified natural
characteristics in scattered regions of mainland China. However, an integral and representative work is
Keywords:
Hydrology
still not available to address potential concerns in general matters. In this study, a large set of rainfall-
Curve number runoff monitoring data were collected to adjust CN values in 55 study sites across China. The results
Surface runoff estimation showed that the revised CN values are largely different from CN look-up table provided by USDA, which
Modeling would lead to huge errors in runoff estimation. In this study, the revised CN (dubbed CN-China) provides
better reference guidelines that are suitable for most natural conditions in China. In addition, scientists
and engineers from other parts of the world can take advantage of the proposed work to enhance the
quality of future programs related to surface runoff estimation.
© 2020 Elsevier Ltd. All rights reserved.

1. Introduction 2008). Surface runoff estimation serves as the most important role in
hydrology related research (Hawkins,1993; Jiang et al., 2012; Kim and
The high-density construction of meteorological stations around Sansalone, 2008; Kirchner et al., 2000; Muche et al., 2019; Steenhuis
the world makes it easy to acquire historical and real-time rainfall et al.,1995; Tyagi et al., 2008; Wang and Wang, 2018). Fortunately, the
data, but there are few stations for runoff monitoring. Runoff moni- relationship between runoff and rainfall allows us to use mathe-
toring is often carried out for short-term and limited area in order to matical methods to estimate runoff based on readily available rainfall
accomplish specific research projects (Fauvel et al., 2016; Gwynne and data (Clyde and Work, 1943; Guo et al., 2017). The Green-Ampt
Glover, 1961; Hvitvedjacobsen and Yousef, 1988; Kim and Sansalone, infiltration curve (Freyberg et al., 1980; Li et al., 2015; Stewart,
2018), the Philip infiltration curve (Triadis and Broadbridge, 2012),
and the Horton infiltration curve methods are all for calculating runoff
* Corresponding author. (Esen, 1987; Grimaldi et al., 2013). Applications of these methods are
** Corresponding author. very limited because they require many parameters and detailed soil
E-mail addresses: [email protected] (Q. Lei), [email protected] (H. Liu).

https://doi.org/10.1016/j.watres.2020.115767
0043-1354/© 2020 Elsevier Ltd. All rights reserved.
2 H. Lian et al. / Water Research 177 (2020) 115767

attribute data that are difficult to acquire. In 1949, Mockus proposed a revised CN values that can be better representative to reflect actual
framework to predict runoff based on land use, soil, antecedent pre- hydrological conditions in China, three objectives were defined to:
cipitation, storm duration, and annual mean temperature based on (i) calculate CN values using the monitored rainfall-runoff data
mechanism of storage-excess flow (Mockus, 1949). The USDA Soil under different climate, soil, land use, and slope conditions across
Conservation Service (SCS) now the Natural Resources Conservation mainland China (contiguous China); (ii) compare CN values ob-
Service (NRCS) proposed CN method in 1954 according to this tained from CN look-up table with the calculated CN values under
framework (Boughton, 1989; Hawkins et al., 2019), which was pub- corresponding field conditions (soil, land use and slope); and, (iii)
lished in the National Engineering Handbook (USDA,1954) in order to identify the key factors differentiating CN values under Chinese
meet the design needs of the Watershed Protection and Flood Pre- field conditions from those obtained by NRCS.
vention Act (Public Law 83-566) (Hawkins et al., 2019). The CN
method has been widely used since then. 2. Materials and methods
The conceptual and empirical CN method is simple with only
one parameter CN, making it the most widely used method for 2.1. Overview of SCS-CN method
calculating runoff based on rainfall events (D’Asaro et al., 2014;
Garen and Moore, 2005; Gaume et al., 2004; Lal et al., 2019; Wilson The CN method was developed by USDA in 1950s originally
et al., 2017). Besides, the CN method has been widely adopted in intended for estimating the depth of surface runoff in an ungauged
many hydrological models (Hawkins et al., 2019; Soulis et al., 2017). small catchment and later implemented also to non-agricultural
The CN method will be performed well if model users obtain CN watersheds with other extended applications (Boughton, 1989;
values from measured rainfall-runoff data when available (Lal et al., Hawkins et al., 2019; Ponce and Hawkins, 1996). The simple
2017). CN parameters reflect the effect of soil infiltration capacity structure approach relies on the parameter CN, a lumped expres-
(Ponce and Hawkins, 1996; Zoure et al., 2019). Key impact factors of sion of a watershed potential hydrologic response. The CN method
CN include soil properties, land use type, slope, antecedent mois- is based on a water balance function (Eq. (1)) and two fundamental
ture conditions, vegetation coverage, land management practices hypotheses. Eq. (1) shows the rainfall depth equals to the summery
(Boughton, 1989, Chin, 2017, Hawkins, 1978a, b, Lal et al., 2019, Shi of Ia, cumulative infiltration (F), and direct runoff (Q). The first
et al., 2017, Zoure et al., 2019). CN values have been derived hypothesis is that the ratio of direct runoff to the maximum po-
experimentally from rainfall-runoff events measurement over a tential runoff is equal to the ratio of the amount of actual infiltration
wide range of geographic, soil, and land management conditions. to the potential maximum retention (Eq. (2)). The second hypoth-
Based on the monitoring data of 150 watersheds in the U.S., sci- esis states that the amount of initial abstraction is some fraction of
entists organized a CN look-up table (USDA, 1954), which is widely the potential maximum retention (Eq. (3)) (Mishra and Singh,
used around the world. Subsequently, Hawkins found that the 1999). The method has interpreted empirical relationships be-
runoff calculated by this method was very sensitive to CN values tween initial abstraction (Ia) and potential maximum retention or
even more than rainfall depth (Hawkins, 1975). Previous study has infiltration (S) through extensive rainfall-runoff experimental data:
shown that a variation of ±10% of CN value leads to a variation Ia ¼ 0.2S. Functions for S (Eq. (4)) and CN (Eq. (5)) are as follow:
from
45% to þ55% runoff (Boughton, 1989). Therefore, the accu-
racy of the CN value plays a crucial role in the estimation of runoff. P ¼ Ia þ F þ Q (1)
Various applications and modifications of CN method have been
implemented in China, including finding the local initial abstrac- Q F
tion (Ia) and CN values based on rainfall-runoff monitoring data (Fu ¼ (2)
P
Ia S
et al., 2011; Jiao et al., 2015; Shi et al., 2009), the effects of different
soil, land use, slope on CN values (Dong et al., 2015; Huang et al.,
Ia ¼ lS (3)
2007; Huang et al., 2006), relationship between CN value and the
rainfall depth under long-term sequence and using remote sensing
 1=2 
to study the spatial heterogeneity of CN values (Chen et al., 2017; S ¼ 5 P þ 2Q
4Q 2 þ 5PQ (4)
Zhang et al., 2019). There is a common conundrum that scientists
could not find the corresponding CN values from CN look-up table.
The primary reasons are the complexity and interactions among CN ¼ 25400=ðS þ 254Þ (5)
land use type, soil infiltration capacity, and climate characteristics
which could result in an enormous error between CN from look-up where, P ¼ rainfall depth (mm); Ia ¼ initial abstraction of the
table and reality. For instance, nearly 70% of the rainfall in the U.S. rainfall (mm); F ¼ cumulative infiltration excluding Ia (mm);
will seep into the soil, while farmlands are fragmented and densely Q ¼ runoff depth (mm); S ¼ potential maximum retention or
drained in China. Infiltration rate is low due to the dense surface infiltration (mm); and, l is the initial abstraction coefficient,
irrigation and drainage system (He et al., 2001). In addition, topo- empirically l is treated as a constant 0.2, the relationship between
graphic variations in China are substantially more complicated than Ia and S is Ia ¼ lS; CN ¼ curve number, which is a dimensionless
in the U.S. Therefore, parameters are being over-calibrated in order parameter with the range from 0 to 100, and the higher the CN
to achieve better simulation performance, especially when users value, the greater the potential of surface runoff.
adopt CN values from look-up table to hydrological models. The The SCS model has no strict physical theory, and its runoff
embedded uncertainty with potentially large errors in surface calculation formula is summarized by measured data from more
runoff estimation is going to affect the corresponding scientific than 150 small watersheds in the U.S. It reflects the empirical law of
credibility in the following analytical processes such as the simu- rainfall-runoff events. The key to affect the accuracy of results is
lation of nitrogen and phosphorus in watershed scale (Yen et al., whether the parameter CN value can accurately reflect the water-
2014). Thus, a national-scale research is urgently needed to alle- shed characteristics. The range of CN values is 0e100. In this study,
viate the tremendous differences in CN value and the correspond- spatial variable CN values are calculated into a single value to
ing predominant controlling factors. represent the average condition of study sites. The arithmetic
The primary goal of this study was to adjust CN values based on means is used to determine the CN value. When CN values are
the monitored rainfall-runoff data in China. In order to provide calculated from real storm data as outlined in the preceding
 H. Lian et al. / Water Research 177 (2020) 115767 3

section, a secondary relationship always happened between 2.3. Validation of the revised CN
decreased CN and increased rainfall depth, and finally CN approach
a constant (Hawkins, 1993). There are many methods to determine In order to verify whether CN values calculated based on the
CN values from observed rainfall-runoff data, while no agreement monitored data can well simulate the runoff in the study area, four
has been settled because no one method shows outstanding research sites were selected with abundant data and divided the
advantage (Lal et al., 2017). In addition, the fact that we use mean rainfall event into two parts. The first part was used to determine
CN value otherwise asymptotic CN value is the lack of abundant CN values, and the second part was used to verify the validity of the
rainfall-runoff data at each study site. What’s more, frequency CN value by using the calculated CN value to estimate the runoff
matching is used to calculate the chosen return-period runoff from depth and compare with the measured runoff depth. The Nash-
the same return-period rainfall depth (Hawkins, 1993). The rainfall Sutcliffe Efficiency (NSE) (Nash and Sutcliffe, 1970) was used to
and runoff depths are collected separately and then arranged on a evaluate the performance of the method. The NSE was used as
rank-order to produce rainfall-runoff pairs during equal return indices of the agreement between the calculated and observed
periods (Hawkins et al., 2009). In this work, we used the actual values of runoff depth (perfect fit when NSE equals to 1, and the
rainfall-runoff events data to calculate CN values other than the worse value of NSE is -∞). When the Nash coefficient is above 0.6,
rank-order data, because we don’t have enough rainfall-runoff the calculated values were deemed qualified.
events in some study sites to match rainfall-runoff data in the
PN
same return period. Therefore, no obvious difference identified ðQ
Qc Þ2i
NSE ¼ 1
P i¼1 0 (6)
between the two forms of rainfall-runoff data. N 2
i¼1 ðQ0
Q0 Þi

where, Qo ¼ observed runoff depth; Qc ¼ calculated runoff depth.

2.2. Data collection

2.4. CN values obtained from the table CN entries
In this study, rainfall-runoff monitoring data were collected
from 31 monitoring plots and 24 watersheds to calculate the CN The CN values for these 55 study sites also obtained from the
value. These sites located in 21 provinces in China. The distribution lookup table according to land use type, management practice,
of all study sites is shown in Fig. 1. Total 55 sites were selected hydrological condition (HC) and hydrologic soil groups (HSG). HSG
depending on available monitoring rainfall-runoff event data. The (A/B/C/D) is classified according to soil saturated hydraulic con-
land use, soil and plant type, slope, and other attributes for these ductivity Ks (mm/h). A (Ks >180), B (18< Ks 180), C (1.8< Ks 18), D
sites are showed in Table 1. In addition, the detailed data sources for (Ks <1.8). Ks is calculated according to the soil properties data
each study site are showed in Supplementary Material Table S2. The (Table 1) from the Soil Testing and Formula Fertilization Program.
soil and land use data are derived from soil surveys conducted by The equation for calculating Ks is as follows:
Soil Testing and Formula Fertilization Program, while the slope and
climatic data are collected from literature review. The rainfall- Ks ¼ 0:056C þ 0:016s þ 0:231Om
0:693 (7)
runoff data used to determine CN values were collected from
literature and Chinese Ecosystem Research Network (CERN). The where, C is the percentage of clay content in soil; s is the percentage
time scale of rainfall-runoff data is the day. of sand in soil, and Om is the percentage of organic matter in soil.

Fig. 1. The distribution of monitoring sites for rainfall and runoff. There are totally 55 sites, including 24 for watershed scale and 31 for plot scale.
Table 1

4
Properties for all 55 study sites. Including the number of rainfall-runoff events N, annual rainfall (mm), altitude (m), soil clay content (%), soil organic matter (SOM) (%), Ks, HSG (hydrological soil group), land use type and slope ().

ID Province Watershed Area/ Plot Area/ Time N Annual Rainfall/ Altitude/ Clay Sand SOM Ks (mm/ HSG CN CN Looked Slope Lon Lat Land use
km2 m2 mm m % % % h) Calculated up

1 Sichuan 100.00 2011 371 2033 3850 12.82 56.15 2.84 158.58 B 94.73 58 NA 101.96 29.99 alpine meadow
e2014
2 Beijing 3.22 2004 42 669 660 6.13 62.78 7.44 52.36 B 97.45 63 NA 117.18 40.58 multiple land use type
3 Yunnan 100.00 2011 115 1500 670 66.18 24.89 0.94 232.99 A 77.14 45 NA 100.68 21.59 broadleaf evergreen forest
e2014
4 Jiangsu 100.00 2004 42 1030 50 43.34 19.91 2.04 162.75 B 90.54 60 NA 120.60 28.10 broadleaf evergreen forest
e2006
5 Jiangxi 100.00 2014 60 1465 70 14.64 55.84 2.86 93.67 B 88.43 60 10 114.92 25.85 forest
6 Guangdong 100.00 2011 136 1613 NA 56.89 17.89 2.34 184.31 A 68.86 45 NA 113.33 23.33 broadleaf evergreen & deciduous scrub
e2014
7 Guizhou 1441.00 2003 5 1220 670 21.89 58.03 1.27 121.01 B 82.12 60 NA 107.29 26.09 multiple land use type
8 Guangdong 25.00 2002 5 1523 40 8.60 84.78 2.40 80.03 B 87.34 66 2 110.30 21.20 forest
9 Beijing 176.00 1982 1 537 NA 13.20 63.34 2.14 93.26 B 77.60 58 NA 117.07 40.53 multiple land use type
e2011
10 Heilongjiang 100.00 2000 36 455 176 44.89 28.54 0.82 176.00 B 85.41 66 10 123.48 47.90 pine forest
11 Beijing 50.00 1993 3 484 660 6.13 62.78 0.94 52.36 B 97.20 78 14.6 117.18 40.58 dryland
e2006
128.10 42.40 broadleaf deciduous & needleleaf evergreen

H. Lian et al. / Water Research 177 (2020) 115767

12 Jilin 100.00 2011 67 800 1090 26.06 30.94 4.06 132.00 B 84.89 66 NA
e2014 forest
13 Beijing 59.10 2006 20 473 1320 22.21 48.50 5.13 150.72 B 78.24 60 8 116.34 40.28 multiple land use type
̊
14 Fujian 100.00 2011 33 1478 5 15.12 31.58 1.79 64.37 B 83.73 66 15 119.14 26.16 orchard
15 Hubei 16.70 1989 21 1013 250 27.04 49.45 0.96 109.98 B 76.71 60 20 110.46 31.13 forest
e1996
16 Hunan 5600.00 1990 28 1474 150 27.93 29.15 3.82 133.13 B 86.59 72 NA 112.72 25.73 multiple land use type
e2009
17 Guangdong 100.00 2011 201 1737 NA 3.15 89.95 0.25 58.84 B 80.27 66 NA 112.52 23.10 broadleaf evergreen & deciduous scrub
e2014
18 Shaanxi 3.70 1958 8 500 1300 7.86 73.66 0.47 62.04 B 71.88 58 30 110.32 39.12 montane steppe
e1966
19 Shaanxi 2.00 1958 18 500 1200 13.47 55.65 1.81 82.24 B 94.60 81 15 109.30 36.80 dryland
e1966
20 Sichuan 100.00 2006 42 863 1800 41.03 31.75 1.23 143.82 B 90.26 77 5 102.18 27.54 dryland
21 Shanxi 1146.00 1982 4 510 1525 7.79 78.86 0.36 65.27 B 91.00 78 NA 112.00 39.03 multiple land use type
e1987
22 Beijing 158.00 1981 1 537 NA 13.20 63.34 2.14 93.26 B 72.89 60 NA 117.07 40.53 multiple land use type
e1994
23 Zhejiang 1444.00 2013 5 1425 650 18.98 57.59 0.66 108.21 B 87.88 76 NA 121.16 30.04 multiple land use type
24 Beijing 50.00 2001 9 508 NA 24.71 43.52 1.40 92.42 B 88.68 78 NA 116.04 40.59 dryland
e2006
25 Guangdong 100.00 1995 13 1720 NA 7.24 60.44 0.75 60.15 B 75.63 66 NA 113.37 23.07 orchard
26 Ningxia 100.00 1993 16 428 3940 9.21 62.06 0.81 90.73 B 94.75 86 10 105.30 36.00 bare land
27 Anhui 50.00 2015 6 995 30 36.35 26.11 0.89 116.06 B 87.85 81 NA 117.55 31.75 dryland
28 Shaanxi 0.50 2015 16 635 580 21.98 26.59 2.22 69.01 B 86.90 81 10 108.04 34.17 dryland
29 Hubei 0.14 2016 12 850 35 28.78 24.40 3.02 90.92 B 90.70 86 5 117.15 31.70 bare land
30 Shaanxi 3.39 1959 3 458 477 19.66 38.10 0.72 70.35 B 80.85 78 3 109.17 34.76 dryland
e2005
31 Shandong 10336.00 1997 28 671 NA 27.21 53.00 1.08 110.71 B 83.37 81 NA 118.50 36.42 multiple land use type
32 Sichuan 100.00 2013 25 826 570 13.17 67.92 0.74 82.86 B 76.29 74 6.5 105.32 31.49 cropland
33 Shandong 323.10 2004 11 612 NA 45.49 30.29 2.12 150.60 B 82.82 81 NA 117.05 36.42 multiple land use type
e2005
34 Beijing 100.00 2011 35 626 1300 20.73 41.92 0.61 97.75 B 67.71 66 NA 115.31 40.11 broadleaf deciduous scrub
e2014
35 Anhui 2670.00 1982 17 1600 100 20.49 47.11 3.36 81.00 B 75.47 75 NA 118.40 29.76 multiple land use type
36 Guizhou 100.00 2012 5 863 1380 47.34 20.06 0.78 183.36 A 44.40 45 5 105.43 27.00 broadleaf evergreen & deciduous scrub
e2014
37 Gansu 8.48 1989 5 300 1970 12.15 53.92 0.28 61.86 B 73.35 74 35 104.26 35.62 dryland
38 Jiangxi 100.00 24 1881 NA 18.75 68.25 1.97 90.81 B 77.27 78 NA 116.77 28.34 Subtropical park woodland
 H. Lian et al. / Water Research 177 (2020) 115767 5

3. Results

101.20 26.58 broadleaf deciduous & needleleaf evergreen

109.71 26.86 broadleaf deciduous & needleleaf evergreen

3.1. Relationships between rainfall and runoff events

104.02 31.67 broadleaf evergreen & deciduous scrub

There is a linear correlation between rainfall and runoff data
from 401 events (Fig. 2). Distribution of scatters in the graph is

109.71 26.86 needleleaf evergreen forest

highly concentrated. Slope of the trend line in the graph reflects
100.18 25.35 multiple land use type

multiple land use type

the runoff coefficient, which is around 0.62. The intercept repre-
sents the minimum rainfall that produces runoff, which is
20.94 mm. It was found that the SCS-CN model has better simu-
lation effect on study areas with runoff coefficient greater than 0.5

cropland
than areas with runoff coefficient less than 0.5 (Peng and You,
105.57 31.06 dryland
108.19 35.21 dryland

106.06 34.65 dryland

109.73 34.91 dryland

110.22 39.23 dryland

dryland

dryland
110.32 39.12 pasture

fallow
forest

forest

forest
2006).
In different study areas with same rainfall depth, the amount

30.07
38.26
30.07
30.84
40.97
26.98
of runoff produced was substantially different. In the same study
107.38 area, the same amount of rainfall at different times or seasons also
110.33
107.38
116.36
122.27
112.39
had a large difference in runoff generation. These two phenomena
indicated that influencing factors of the CN value are complex.
Interactions in external natural geographical conditions will cause
NA

NA

NA
NA

NA

NA

NA

NA

NA

NA
6.5

30

40

41

10

10

10
the difference in regional runoff generation ability, while the in-
ternal seasonal variation and rainfall duration will also cause the
difference in the runoff.
Relationships between CN values and runoff depth are shown
60
78
78

77
60

66

78

69

55

55

78

71
78
86
78
75
60

in Fig. 3 under different rainfall depths. According to the SCS-CN

method, larger CN values may generate greater runoff under the
same rainfall event. However, small rainfall events can still
58.78
76.29
76.01

74.69
57.68

63.06

74.03

64.07

49.47

48.89

70.74

53.03
58.80
64.16
53.23
40.92
12.80

generate runoff by proper amount of initial abstraction and the

soil infiltration ability. It indicates the research area has a strong
ability to produce runoff. Therefore, the corresponding CN value
B
B
B

B
B

B
B
B
B
B
B

could be large.
35.74 28.60 1.01 127.23

36.04 19.61 3.17 142.31

36.04 19.61 3.17 142.31

18.26 55.21 2.18 103.05

102.78

102.78

115.27
124.49
21.32 35.27 1.24 91.27
13.39 63.71 1.03 81.75
28.85 24.94 0.93 93.59

20.71 35.72 2.93 75.15

13.10 46.77 1.53 87.93

20.10 34.33 0.67 72.92

73.66 0.47 62.04

58.47

94.07

3.2. Comparison of calculated and table CN entries CN values

0.34
1.10
1.10
1.42
1.76
1.25

CN values of the 55 study sites were calculated from the

23.35
77.00
23.35
68.83
50.03
44.13

measured data of rainfall and runoff events, and the average CN

value of different events in the same study area represents the CN
31.77

31.77
13.43
27.24
30.97
7.86

6.37

parameters of this site. The calculated CN values have showed

clear differences compared to the CN values obtained from the
Note: the references for these 55 study sites are added in Supplementary Material Table S2.

table CN entries (USDA, 1954) for the study area (Fig. 4, Supple-
1997

1230

1625
2300

1300

1250
965

410

430

430

519
900
519
910

110

mentary materials Table S1). Results revealed that the calculated

NA

10

CN value ranges 12.80e97.45, while the looked-up table indicates

as 36e91. The difference between calculated and looked-up
values ranges between
40.41e34.08. Comparisons of average
value and error at each point indicated that the CN value acquired
1200

1335

1335

1071

1071
1400

1500

by table CN entries have reduced its variation. It could be sub-

547
900
556

522

909

561

440

440

440

480

stantially different by regions, which largely ignores the study

area characteristics. Moreover, the CN table CN entries provided
15

14
90

38

52

51

60

12

18
10
3

by USDA-SCS were summarized from a limited set of watershed

observations (Bartlett et al., 2016; Ogden et al., 2017). Thus, these
e2006

e1995

e2014

e2014

e2008

e1964

e2014

e2014

e1972
2004

2013
2013
1959

1987
2011

2011

1959

1959

2011

2011

1959

2006
1959
2006
1987
2008
2009

differences have supported our doubts on the applicability of the

existing table CN entries in China, and for the areas with different
characteristics in different regions, the existing looked-up table
100.00

100.00

100.00

100.00

100.00

100.00

100.00
60.00

for CN is not suitable for runoff estimation.

5.30

The blue line represents the calculated value of the monitoring

point CN, and the orange line represents the CN value obtained by
the existing look- up the table (Fig. 4). It can be found that CN
value based on the measured data has a large variation range.
Differences between the monitoring points can be clearly found,
219.00

984.00
0.41

1.14

0.33

2.54

0.86

2.54

while the orange line has a small fluctuation range, floating be-
tween 50e80. In literature, it is nearly impossible to see that the
Chongqing

Chongqing

CN value is less than 30 under actual environmental conditions,

Liaoning
40 Sichuan
41 Shaanxi

44 Sichuan

45 Shaanxi

46 Shaanxi

49 Shaanxi

Shaanxi
39 Yunnan

43 Yunnan

and the range of variation is between 30e100. Here, some study

47 Hunan

48 Hunan

Hunan
42 Gansu

Anhui

sites with CN values more than 80 and some are less than 30. After
repeated inspections of the data, the occurrence of extreme CN
50
51
52
53
54
55

values is because of the different forest type and local characters.

6 H. Lian et al. / Water Research 177 (2020) 115767

3.3. Model performance

The purpose of validation conducted in the study was to ensure

the driven results are reasonable in scientific measures. One should
keep in mind that the derived results could be biased by changing
validation standard (Yen et al., 2016). In this study, 0.6 is the value
of NSE set to be the statistical guideline of satisfactory. To figure out
the reliability of CN method, the simulated runoff was compared
with observed runoff depths. For this purpose, four sites with large
set of rainfall-runoff data were selected to validate the performance
of this model. The freedom of rainfall-runoff events to validate the
accuracy of CN parameters of these four sites is 7, 14, 8, and 14
respectively. The NSE results are 0.98, 0.76, 0.86, and 0.84 respec-
tively, far more than 0.6, which indicates that the CN method is
accurate enough to simulate runoff depth (Fig. 5).

4. Discussions

4.1. Differences of CN values between the United States and China

Studies have shown that soil moisture contents, hydrological

processes, infiltration, and other losses through rainfall have sig-
nificant spatial and temporal variability (Ponce and Hawkins, 1996; Fig. 3. Relationships between CN values and runoff depth under different rainfall
Tramblay et al., 2010; Wang, 2018; Zeng et al., 2017). It indicated depths.
that the forestland has lowest runoff production capacity and the
paddy field has highest (Choi et al., 2019). Other researchers also
found that the land use types with high CN values are mainly urban 4.2. Impact factors of the revised CN
area, farmland and water bodies, and the low CN values are in
forests, orchards, shrubs and grasslands (Choi et al., 2019; Failure to evaluate the size of regional scale might affect the
Egodawatta et al., 2007; Li et al., 2018; Soulis et al., 2017; Zhang corresponding error between actual runoff and CN parameters
et al., 2018). This might be associated with less vegetation cover simulated runoff. By summarizing the previous studies, it was
in urban area, farmland and water bodies compared to forests. found that this model is suitable for the study of small and
Although CN-method may introduce large uncertainty in runoff medium-scale watersheds, with an area of about 0.25e1000 km2.
estimation of paddy field, many scholars still treat paddy filed as Even the CN method was appropriate still the dominant behavior of
one kind of land use type with low infiltration capacity in order to CN value’s response to rainfall depth is important to calculate the
estimate runoff depth easily. The main character of curve number runoff depth (Muche et al., 2019; Soulis et al., 2009). Results indi-
method is its base on accumulated experience in use, mainly in the cated a decrease in CN value with increase in rainfall but stable at
U.S. There are many weaknesses and limitations when the original high rainfall intensity (Fig. 6). It indicates that the rainfall depth is
CN values were applied in China (Boughton, 1989). Furthermore, the primary factor to determine the CN value, and the asymptotic
some watersheds performed quite differently from the basic CN constant value is the best way to identify CN parameters for a
runoff response patterns, leading to great differences between watershed. And other studies have showed that CN values calcu-
model and the actual processes. The major weakness and utmost lated from measured rainfall-runoff data vary systematically with
potential source of error is the sensitivity of estimated runoff in the the rainfall depth, the determination of a single asymptotic CN
selection of the curve number. value observed for very high rainfall depths to characterize the
watershed’s runoff response (Soulis and Valiantzas, 2012). Standard
behavior of CN response for the rainfall-runoff relationship found in
our study with a distinct bias for high CN at lower rainfall depths
indicates that rainfall is another key factor in determining CN.
Rainfall duration could be an important phenomenon to in-
crease soil water content, which will be at maximum after certain
rainfall period (Thorndahl and Willems, 2008). After attaining
maximum soil water contents, CN values will be increased nearly
up to 100 indicating runoff generation capacity is also increased. As
the previous soil moisture condition (AMC) has a great influence on
the results (Lal et al., 2019), the study sites with detailed monitoring
data can be more reliable. For same test sites, runoff coefficient is
different under different rain intensity. Therefore, CN value as the
parameter of regional runoff needs to be able to represent the
runoff property of the region, which needs to be calculated. Long-
term monitoring of rainfall in different seasons and intensity can
reflect the overall situation of the area (Yeh et al., 2018).
Curve number varies spatially with the change in land use type,
soil type and soil humidity in the early stage (Ross et al., 2018).
Fig. 2. Relationships between rainfall and runoff of all monitoring sites. A total of 600 Urban land and water bodies have higher CN value compared to
rainfall events were collected across the country.
 H. Lian et al. / Water Research 177 (2020) 115767 7

Fig. 4. The calculated CN value and looked-up CN value for 55 study sites. Only one third of total study sites can use table CN entries (USDA and Soil Conservation Service 1954) from
USDA to gain satisfied (±5%) CN values.

Fig. 5. The results of SCS-CN method validation. Four study sites with enough data are selected to validate the accuracy of the model. Qob is the observed runoff depth, Qca is the
calculated runoff depth, and n represent the number of rainfall-runoff events.

forests, mountains, and hills (Yang and Toor, 2017). The effect of soil detailed hydrological conditions for some study sites. In this study,
properties on CN is divided into four categories according to the we have tried our best to collect more data in order to cover more
permeability of soil. American soil experts initially based on more regions of China and collect more rainfall-runoff events to calculate
than 14,000 soil data classified soils with similar runoff generation accurate results.
capacity called the hydrological soil group (HSG) (Ross et al., 2018; The runoff calculated by the SCS-CN method is known to be
Stewart et al., 2012). The runoff generation capacities from low to sensitive to CN values, which is the largest potential source of error.
high are divided into four categories: A, B, C, and D. Detailed clas- Low frequency and poor resolution of monitoring data will result in
sification criteria showed in the US National Engineering Hand- poor estimation of runoff. Even with years of data available for
book, 1972 (USDA, 1972). calibration of the curve numbers, substantial errors may still occur
(Boughton, 1989). Obviously, we are getting more and more accu-
4.3. Shortcomings and future perspectives rate and high-resolution rainfall data through advanced observa-
tion devices. Thus, a strong runoff estimation method is urgently
The main limitation of our study is the ignorance of antecedent needed. While it is very difficult to establish the mechanism
moisture condition (AMC). Because the rainfall-runoff event data equation of rainfall and runoff because infiltration capacity is
we used did not have the rainfall data of the days before the event affected by many factors such as soil properties, vegetation
occurred, it was impossible to know the soil moisture conditions in coverage, land use type, climate, etc. Fortunately, we have such a
the early stage of the event. Another limitation is the number of simple empirical method to estimate runoff and meet the needs of
rainfall-runoff events is still not large enough to reflect the more massive simulations. There are several studies mapping the global
8 H. Lian et al. / Water Research 177 (2020) 115767

Acknowledgments

This study was supported by funding from the National Natural

Science Foundation of China (Grant No.: 31572208); the Special
Fund for Agro-scientific Research in the Public Interest (Grant No.:
201303089); and the Newton Fund (Grant Ref: BB/N013484/1). And
acknowledgment for the data support from Chinese Ecosystem
Research Network (CERN).

Appendix A. Supplementary data

Supplementary data to this article can be found online at

https://doi.org/10.1016/j.watres.2020.115767.

References

Bartlett, M.S., Parolari, A.J., McDonnell, J.J., Porporato, A., 2016. Beyond the SCS-CN
Fig. 6. Response of CN values to rainfall intensity. Decreasing CN value with increasing method: a theoretical framework for spatially lumped rainfall-runoff response.
rainfall but stable at high rainfall intensity. Water Resour. Res. 52 (6), 4608e4627.
Boughton, W.C., 1989. A review of the USDA SCS curve number method. Aust. J. Soil
Res. 27 (3), 511e523.
Chen, H.L., Liang, Z.Y., Liu, Y., Liang, Q.H., Xie, S.G., 2017. Integrated remote sensing
CN values based on Moderate Resolution Imaging Spectroradi- imagery and two-dimensional hydraulic modeling approach for impact evalu-
ometer (MODIS), Harmonized World Soil Database (HWSD), and ation of flood on crop yields. J. Hydrol. 553, 262e275.
Normalized Difference Vegetation Index (NDVI) data sets (Lin et al., Chin, D.A., 2017. Estimating the parameters of the curve number model. J. Hydrol.
Eng. 22 (6), 06017001.
2017; Zeng et al., 2017). Those methods are efficient to produce Choi, D., Park, H., Kim, Y.J., Jung, J.W., Choi, W.J., Her, Y.G., Yoon, K.S., 2019. Curve
high-resolution CN values. While the huge error will be introduced numbers for rice paddies with different water management practices in korea.
by the original CN lookup table which may not suitable for other J. Irrigat. Drain. Eng. 145 (5), 06019003.
Clyde, G.D., Work, R.A., 1943. Precipitation runoff relationships as a basis for water-
countries. Our work is going to be very useful to resolve this issue supply forecasting. Trans. Am. Geophys. Union 24, 43e49.
with relatively minor effort. And future research should focus on D’Asaro, F., Grillone, G., Hawkins, R.H., 2014. Curve number: empirical evaluation
the establishment of a dynamic runoff estimation model. The input and comparison with curve number handbook tables in sicily. J. Hydrol. Eng. 19
(12), 04014035.
data of the model should include meteorological, topographical, Dong, L.H., Xiong, L.H., Lall, U., Wang, J.W., 2015. The effects of land use change and
soil type, land use, and vegetation cover data sets. precipitation change on direct runoff in Wei River watershed, China. Water Sci.
For ungauged watersheds, CN parameters are usually estimated Technol. 71 (2), 289e295.
Egodawatta, P., Thomas, E., Goonetilleke, A., 2007. Mathematical interpretation of
by well-known handbook tables (USDA, 1972), according to the
pollutant wash-off from urban road surfaces using simulated rainfall. Water
hydrologic soil group (HSG), land use, surface condition and ante- Res. 41 (13), 3025e3031.
cedent moisture condition (AMC). In the presence of variable Esen, I.I., 1987. Least-squares estimates of the Horton infiltration parameters. Soil
conditions (soil type, land cover, and land use) within the basin, an Sci. 144 (1), 6e10.
Fauvel, B., Cauchie, H.M., Gantzer, C., Ogorzaly, L., 2016. Contribution of hydrological
area-weighted average CN is often used. According to our results, it data to the understanding of the spatio-temporal dynamics of F-specific RNA
is urgently needed a CN handbook table for different regions, which bacteriophages in river water during rainfall-runoff events. Water Res. 94,
can be established based on local rainfall runoff monitoring 328e340.
Freyberg, D.L., Reeder, J.W., Franzini, J.B., Remson, I., 1980. Application of the green-
network data. ampt model to infiltration under time-dependent surface-water depths. Water
Resour. Res. 16 (3), 517e528.
Fu, S., Zhang, G., Wang, N., Luo, L., 2011. Initial abstraction ratio in the scs-cn method
5. Conclusion in the loess plateau of China. Transact. Asabe 54 (1), 163e169.
Garen, D.C., Moore, D.S., 2005. Curve number hydrology in water quality modeling:
uses, abuses, and future directions. J. Am. Water Resour. Assoc. 41 (2), 377e388.
The Curve Number approach is an effective, simple and widely Gaume, E., Livet, M., Desbordes, M., Villeneuve, J.P., 2004. Hydrological analysis of
used method to estimate surface runoff volume. The revised CN the river Aude, France, flash flood on 12 and 13 November 1999. J. Hydrol. 286
(1e4), 135e154.
values calculated by using monitoring data from 55 sites in China Grimaldi, S., Petroselli, A., Romano, N., 2013. Green-Ampt Curve-Number mixed
are considerably different from the CN values acquired from US- procedure as an empirical tool for rainfall-runoff modelling in small and
based CN value lookup-table. It has been shown that the moni- ungauged basins. Hydrol. Process. 27 (8), 1253e1264.
Guo, A.J., Chang, J.X., Liu, D.F., Wang, Y.M., Huang, Q., Li, Y.Y., 2017. Variations in the
toring data can provide valuable information to calculate the CN precipitation-runoff relationship of the weihe river basin. Nord. Hydrol 48 (1),
value. In this study, the modified CN values (CN-China) derived 295e310.
from the actual rainfall-runoff data play an important role and a Gwynne, M.D., Glover, J., 1961. Light rainfall and plant survival - meaurement of
stem flow run-off. Nature 191 (479), 1321. &.
very useful reference in future runoff estimation. The results of this
Hawkins, R.H., 1975. The importance of accurate curve numbers in the estimation of
study provide new insight into the establishment of the CN table storm runoff. Water Resour. Bull. 11 (5), 887e891.
that is more suitable for China and possibly many other countries Hawkins, R.H., 1978. Infiltration formula based on scs curve number. J. Irrigation
and Drainage Division-Asce 104 (4), 464e467.
around the world. One can take advantage of the given work to
Hawkins, R.H., 1978. Runoff curve numbers with varying site moisture. J. Irrigation
incorporate more details of the curve number approach to further and Drainage Division-Asce 104 (4), 389e398.
improve the corresponding accuracy and applicability. Hawkins, R.H., 1993. Asymptotic determination of runoff curve numbers from data.
J. Irrigat. Drain. Eng. 119 (2), 334e345.
Hawkins, R.H., Theurer, F.D., Rezaeianzadeh, M., 2019. Understanding the basis of
the curve number method for watershed models and TMDLs. J. Hydrol. Eng. 24
Declaration of competing interest (7), 06019003.
Hawkins T.W., R.H., Ward, T.J., Woodward, D.E., Van Mullem, J.A., 2009. Curve
The authors declare that they have no known competing Number Hydrology State of the Practice. American Society of Civil Engineers,
Reston Va, 106pp.
financial interests or personal relationships that could have He, B., Zhou, N., Gao, X., 2001. Precipitation-runoff relationship in farmland
appeared to influence the work reported in this paper. nonpoint source pollution research: amending coefficient of SCS hydrologic
 H. Lian et al. / Water Research 177 (2020) 115767 9

method. Res. Environ. Sci. 14 (3), 49e51. Soulis, K.X., Valiantzas, J.D., Dercas, N., Londra, P.A., 2009. Investigation of the direct
Huang, M.B., Gallichand, J., Wang, Z.L., Goulet, M., 2006. A modification to the Soil runoff generation mechanism for the analysis of the SCS-CN method applica-
Conservation Service curve number method for steep slopes in the Loess bility to a partial area experimental watershed. Hydrol. Earth Syst. Sci. 13 (5),
Plateau of China. Hydrol. Process. 20 (3), 579e589. 605e615.
Huang, M.B., Gallichand, J., Dong, C.Y., Wang, Z.L., Shao, M.G., 2007. Use of soil Soulis, K.X., Ntoulas, N., Nektarios, P.A., Kargas, G., 2017. Runoff reduction from
moisture data and curve number method for estimating runoff in the Loess extensive green roofs having different substrate depth and plant cover. Ecol.
Plateau of China. Hydrol. Process. 21 (11), 1471e1481. Eng. 102, 80e89.
Hvitvedjacobsen, T., Yousef, Y.A., 1988. Analysis of rainfall series in the design of Steenhuis, T.S., Winchell, M., Rossing, J., Zollweg, J.A., Walter, M.F., 1995. Scs runoff
urban drainage control-systems. Water Res. 22 (4), 491e496. equation revisited for variable-source runoff areas. J. Irrigation Drain. Eng. Asce
Jiang, W.Y., Haver, D., Rust, M., Gan, J., 2012. Runoff of pyrethroid insecticides from 121 (3), 234e238.
concrete surfaces following simulated and natural rainfalls. Water Res. 46 (3), Stewart, R.D., 2018. A dynamic multidomain green-ampt infiltration model. Water
645e652. Resour. Res. 54 (9), 6844e6859.
Jiao, P.J., Xu, D., Wang, S.L., Yu, Y.D., Han, S.J., 2015. Improved SCS-CN method based Stewart, D., Canfield, E., Hawkins, R., 2012. Curve number determination methods
on storage and depletion of antecedent daily precipitation. Water Resour. and uncertainty in hydrologic soil groups from semiarid watershed data.
Manag. 29 (13), 4753e4765. J. Hydrol. Eng. 17 (11), 1180e1187.
Kim, J.Y., Sansalone, J.J., 2008. Event-based size distributions of particulate matter Thorndahl, S., Willems, P., 2008. Probabilistic modelling of overflow, surcharge and
transported during urban rainfall-runoff events. Water Res. 42 (10e11), flooding in urban drainage using the first-order reliability method and
2756e2768. parameterization of local rain series. Water Res. 42 (1e2), 455e466.
Kirchner, J.W., Feng, X.H., Neal, C., 2000. Fractal stream chemistry and its implica- Tramblay, Y., Bouvier, C., Martin, C., Didon-Lescot, J.F., Todorovik, D., Domergue, J.M.,
tions for contaminant transport in catchments. Nature 403 (6769), 524e527. 2010. Assessment of initial soil moisture conditions for event-based rainfall-
Lal, M., Mishra, S.K., Pandey, A., Pandey, R.P., Meena, P.K., Chaudhary, A., Jha, R.K., runoff modelling. J. Hydrol. 387 (3e4), 176e187.
Shreevastava, A.K., Kumar, Y., 2017. Evaluation of the Soil Conservation Service Triadis, D., Broadbridge, P., 2012. The Green-Ampt limit with reference to infiltra-
curve number methodology using data from agricultural plots. Hydrogeol. J. 25 tion coefficients. Water Resour. Res. 48. W07515.
(1), 151e167. Tyagi, J.V., Mishra, S.K., Singh, R., Singh, V.P., 2008. SCS-CN based time-distributed
Lal, M., Mishra, S.K., Kumar, M., 2019. Reverification of antecedent moisture con- sediment yield model. J. Hydrol. 352 (3e4), 388e403.
dition dependent runoff curve number formulae using experimental data of USDA, Soil Conservation Service (SCS), 1954. National Engineering Handbook.
Indian watersheds. Catena 173, 48e58. Section 4. Hydrology. U.S Development of agriculture, Washington, D.C.
Li, J., Wang, Z.G., Liu, C.M., 2015. A combined rainfall infiltration model based on USDA, Soil Conservation Service (SCS), 1972. National Engineering Handbook. Sec-
Green-Ampt and SCS-curve number. Hydrol. Process. 29 (11), 2628e2634. tion 4. Hydrology. U.S Development of agriculture, Washington, D.C.
Li, C.L., Liu, M., Hu, Y.M., Shi, T., Zong, M., Walter, M.T., 2018. Assessing the impact of Wang, D.B., 2018. A new probability density function for spatial distribution of soil
urbanization on direct runoff using improved composite CN method in a large water storage capacity leads to the SCS curve number method. Hydrol. Earth
urban area. Int. J. Environ. Res. Publ. Health 15 (4), 775. Syst. Sci. 22 (12), 6567e6578.
Lin, W., Yang, F., Zhou, L., Xu, J.G., Zhang, X.Q., 2017. Using modified Soil Conser- Wang, S., Wang, H., 2018. Extending the Rational Method for assessing and devel-
vation Service curve number method to simulate the role of forest in flood oping sustainable urban drainage systems. Water Res. 144, 112e125.
control in the upper reach of the Tingjiang River in China. J. Mt. Sci. 14 (1), 1e14. Wilson, L.E., Ramirez-Avila, J.J., Hawkins, R.H., 2017. Runoff Curve Number Estima-
Mishra, S.K., Singh, V.P., 1999. Another look at scs-cn method. J. Hydrol. Eng. 4 (3), tion for Agricultural Systems in the Southern Region of USA. World Environ-
257e264. mental and Water Resources Congress 2017: Watershed Management,
Mockus, V., 1949. Estimation of total (and peak rates of) surface runoff for individual Irrigation and Drainage, and Water Resources Planning and Management,
storms. Exhibit A of Appendix B, Interim Survey Report, Grand (Neosho) River pp. 353e366.
Watershed. USDA Soil Conservation Service. Yang, Y.Y., Toor, G.S., 2017. Sources and mechanisms of nitrate and orthophosphate
Muche, M.E., Hutchinson, S.L., Hutchinson, J.M.S., Johnston, J.M., 2019. Phenology- transport in urban stormwater runoff from residential catchments. Water Res.
adjusted dynamic curve number for improved hydrologic modeling. J. Environ. 112, 176e184.
Manag. 235, 403e413. Yeh, T.C., Liao, C.S., Chen, T.C., Shih, Y.T., Huang, J.C., Zehetner, F., Hein, T., 2018.
Nash, J.E., Sutcliffe, J.V., 1970. River flow forecasting through conceptual models part Differences in N loading affect DOM dynamics during typhoon events in a
I- A discussion of principles. J. Hydrol. 10 (3), 282e290. forested mountainous catchment. Sci. Total Environ. 633, 81e92.
Ogden, F.L., Hawkins, R., Walter, M.T., Goodrich, D.C., 2017. Comment on "Beyond Yen, H., Wang, X.Y., Fontane, D.G., Harmel, R.D., Arabi, M., 2014. A framework for
the SCS-CN method: a theoretical framework for spatially lumped rainfall- propagation of uncertainty contributed by parameterization, input data, model
runoff response. In: Bartlett, M.S. (Ed.), Water Resources Research vol. 53(7), structure, and calibration/validation data in watershed modeling. Environ.
pp. 6345e6350 et al. Model. Software 54, 211e221.
Peng, D., You, J., 2006. Application of modified SCS model into runoff simulation. Yen, H., Hoque, Y.M., Wang, X.Y., Harmel, R.D., 2016. Applications of explicitly
J. Water Res. Water Eng. 17, 20e24. incorporated/post-processing measurement uncertainty in watershed
Ponce, V.M., Hawkins, R.H., 1996. Runoff curve number: has it reached maturity? modeling. J. Am. Water Resour. Assoc. 52 (2), 523e540.
J. Hydrol. Eng. 1 (1), 11e19. Zeng, Z.Y., Tang, G.Q., Hong, Y., Zeng, C., Yang, Y., 2017. Development of an NRCS
Ross, C.W., Prihodko, L., Anchang, J., Kumar, S., Ji, W.J., Hanan, N.P., 2018. HYS- curve number global dataset using the latest geospatial remote sensing data for
OGs250m, global gridded hydrologic soil groups for curve-number-based runoff worldwide hydrologic applications. Remote Sense. Lett. 8 (6), 528e536.
modeling. Scientific Data 5, 150091. Zhang, W., Villarini, G., Vecchi, G.A., Smith, J.A., 2018. Urbanization exacerbated the
Shi, Z.H., Chen, L.D., Fang, N.F., Qin, D.F., Cai, C.F., 2009. Research on the SCS-CN rainfall and flooding caused by hurricane Harvey in Houston. Nature 563,
initial abstraction ratio using rainfall-runoff event analysis in the Three 384e388.
Gorges Area, China. Catena 77 (1), 1e7. Zhang, D.J., Lin, Q.Y., Chen, X.W., Chai, T., 2019. Improved curve number estimation
Shi, W.H., Huang, M.B., Gongadze, K., Wu, L.H., 2017. A modified SCS-CN method in SWAT by reflecting the effect of rainfall intensity on runoff generation. Water
incorporating storm duration and antecedent soil moisture estimation for 11, 163.
runoff prediction. Water Resour. Manag. 31 (5), 1713e1727. Zoure, C., Queloz, P., Koita, M., Niang, D., Fowe, T., Yonaba, R., Consuegra, D.,
Soulis, K.X., Valiantzas, J.D., 2012. SCS-CN parameter determination using rainfall- Yacouba, H., Karambiri, H., 2019. Modelling the water balance on farming
runoff data in heterogeneous watersheds - the two-CN system approach. practices at plot scale: case study of Tougou watershed in Northern Burkina
Hydrol. Earth Syst. Sci. 16 (3), 1001e1015. Faso. Catena 173, 59e70.